Development and evaluation of a sensor enabled smart herbicide applicator for inter row crops

Abstract

A novel self-propelled smart herbicide applicator (1600 mm × 500 mm × 1100 mm) was developed using machine vision to target weeds and apply herbicides variably in wide row-spaced crops. The applicator integrates a weed detection system, smart sprayer, mechanical components, and power transmission system. The weed detection system utilizes a webcam based image sensor, with images processing performed through MATLAB's image processing toolbox. The system processed images using an RGB-based algorithm (R > G + 5, & R > B + 5) to separate weeds from soil. Weed density, calculated as the green area percentage (ratio of green pixels to total pixels), was sent to a microcontroller-operated smart sprayer via a serial port. The microcontroller regulates herbicide flow by controlling the flow control valve with a servo motor, following an algorithm based on green area percentage. Field tests demonstrated a 40 % reduction in herbicide use, 91.26 % weeding efficiency, and 63.4 % field efficiency. The smart herbicide applicator's unique features include real-time measurement, portability, and precise herbicide application for wide-row crops, even at night. It also reduced environmental hazards and drudgery among farmers.

Keywords

• Weed management
• Weed detection
• Machine vision
• Variable rate spraying
• Sensor
• Herbicide applicator